SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Fix metric computation in run_glue_no_trainer (#11569)

Browse Source
This commit is contained in:
Sylvain Gugger
2021-05-03 11:42:55 -04:00
committed by GitHub
parent a721a5eefd
commit 87dd1a00ef
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/pytorch/text-classification/run_glue_no_trainer.py
View File

@@ -404,7 +404,7 @@ def main():
model.eval() model.eval()
for step, batch in enumerate(eval_dataloader): for step, batch in enumerate(eval_dataloader):
outputs = model(**batch) outputs = model(**batch)
predictions = outputs.logits.argmax(dim=-1) predictions = outputs.logits.argmax(dim=-1) if not is_regression else outputs.logits.squeeze()
metric.add_batch( metric.add_batch(
predictions=accelerator.gather(predictions), predictions=accelerator.gather(predictions),
references=accelerator.gather(batch["labels"]), references=accelerator.gather(batch["labels"]),